Databox Generative AI

---

Availability

Users, Editors, and Admins

All accounts

All subscription plans

---

Tracking a metric tells you what happened; deciding whether that's good, bad, or worth acting on takes more effort — usually pulling up historical data, comparing periods, and spotting what changed. Databox's Generative AI does that interpretation for you: it reads the historical data behind your metrics and writes a plain-language summary of what changed, along with recommendations for what to do next, wherever a performance summary is available.

Instead of digging into raw data to understand performance, generative AI surfaces the interpretation directly:

• Understand performance without digging into raw data — a written summary replaces manual comparison across time periods.
• Spot trends and anomalies faster — meaningful changes and outliers are called out instead of buried in a chart.
• Make smarter decisions with less effort — recommendations point to specific areas worth acting on, based on the data analyzed.

Powered by OpenAI, Databox's Generative AI turns raw historical data into a summary in three stages:

1. Data preparation — Databox pulls the relevant metric's history, including the comparison period needed to measure change, and cleans it for consistency.
2. Pattern detection — the model analyzes that data to identify the overall trend, meaningful changes, and any outliers worth flagging.
3. Plain-language output — the analysis is converted into a short written summary plus targeted recommendations, so the interpretation is ready to read rather than something you have to do yourself.

Each generated summary is made up of a performance indicator, a summary paragraph, and up to three recommendations — see Review AI-powered performance summaries for what each part means and how to regenerate them.

• Hovering over a Databoard or metric shows a performance indicator; clicking it generates or opens a summary. See Review AI-powered performance summaries for details.
• Adding the AI Summary Datablock to a Databoard generates a similar summary for up to 10 metrics at once, and lets you edit the result afterward. See Understanding the AI Summary visualization.

AI can surface useful trends and suggestions, but it doesn't replace human judgment. Always review recommendations in the context of your business before acting on them.

Insights are based on the metric data available to Databox and may not reflect factors it can't see, like a seasonal promotion, an internal project, or a shift in the market.

A single spike or outlier can skew a summary's framing. If a summary seems off, check the underlying data source or try a different timeframe before trusting the conclusion.

Because the underlying model learns from historical data, it can carry over patterns and assumptions from that data, including bias. Use the insights to guide your thinking, not dictate it.